Video Technology

MPEG Video Standards

All conventional analogue signals used in audio or video for sound or pictures can be converted to a digital signal. This was first demonstrated for consumer electronics products in compact discs in the early 1980s. In this format, the audio signal is sampled 44100 times every second to a 16-bit digital number and it is this succession of numbers which is recorded on the CD. Later, it was found that it is not necessary to record every single bit in every sample. By removing sounds that are inaudible in the original signal, it is possible to compress the original data so that the signal can be recorded on a medium with a lower capacity than a CD. Two examples of such media are DCC tape and the minidisk.

In spite of the success with audio signals, it had previously been thought that it would be impossible to record video on a medium such as the compact disc. This was primarily because, recording at the full video bandwidth, there was room only for about 24 seconds of a video signal on a conventional CD. Further, the transfer rate of bits from a CD was too slow for sustained video signals. But at a conference in 1989, David Sarnoff Laboratories stunned the word by demonstrating digital video playback from a CD.

Needless to say, this started a chain of research at a number of laboratories to improve the system. This resulted in a number of various compression techniques. A version of the original method which still exists today is the Indeo system, owned by the integrated circuit manufacturer INTEL. Later, major electronics manufacturers got together and worked out a standard to be used for video playback from CDs, which became known as MPEG-1. The name MPEG refers to the standardising group, the so-called Motion Pictures Expert Group. Philips was one of the first to incorporate this standard, which the company launched as a supplement to their CD-i format, and which was initially referred to as full motion video. Later, the name was changed to CD-i Video, and later again, when the format was launched as one independent of CD-i, as Video-CD.

The video signal of each TV frame is converted to digital code in the form of pixels, or a succession of single dots on the TV screen. A description of each dot or pixel is given by the signal level of each of the three colours: red, green and blue. However, in any sequence of TV frames, large areas of the picture remain unchanged and the only changes are small movements that occur from one picture to the next. By storing information only of the changes that have occurred from the previous picture, a large part of the total information does not need to be repeated, and space is saved. In addition, a picture is often composed of large areas where successive pixels have exactly the same colour, for example the area of the sky, of the walls of a building, or areas with water. In these cases, an instruction with the number of pixels with the same data can be recorded together with the data of the pixel. Of course, the exceptions are where there is a change in scene, or to a smaller degree where the camera pans rapidly to follow a moving object.

The MPEG-1 standard only describes how to decode the compressed code used to record the digital video signal, in order that it may be converted to a normal PAL or NTSC television signal. The standard does not include the methods that should be used to compress, or encode the signal. Therefore, the quality of the signal on the CD depends only on the efficiency of the method used to compress the original signal. This is why, although inexpensive programs for home computers exist to compress video to MPEG-1 standards, the results cannot be called high quality. For professional results, a large expensive dedicated machine is required.

The MPEG-1 standard has not only been used for CD-i and Video CD, but also in a few computer programs. However, while the standard can produce quality which is comparable or better than VHS tape, a higher quality standard was required for other applications. The same group therefore designed a new standard, which was named MPEG-2. Because of the higher quality of the signal, the data requirements of MPEG-2 are higher than for MPEG-1, although MPEG-2 is fully compatible with all signals encoded in the MPEG-1 format. Again, the standard only defined the decoding part of the process, leaving the encoding to the ingenuity of the designers of recording equipment.

MPEG-2 has several levels, with different degrees of compression and required rate of transfer of bits. At the highest level, the quality is said to be equal to 35mm film, as used in the cinema. However, the only instance where it is currently used is for the digital satellite TV channels in the USA, but where a bit-rate lower than the highest is used, with a quality similar to MPEG-1. MPEG-2 is the standard that has been adopted for the proposed High Definition television (HDTV) for the USA and will probably also be used for a future digital TV in Europe. It is also the standard for DVD or 'Digital Versatile Disc'. DVDs can store up to 11 times the capacity of a CD on each side of a single disc, making it ideal for storing a full feature film, even with the higher data requirements of MPEG-2 on a single side of the disc.

MPEG-1 and -2 define techniques for compressing digital video by factors varying from 25:1 to 50:1. The compression is achieved using five different compression techniques:

  1. The use of a frequency-based transform called Discrete Cosine Transform (DCT)
  2. Quantization, a technique for losing selective information (sometimes known as lossy compression) that can be acceptably lost from visual information
  3. Huffman coding, a technique of lossless compression that uses code tables based on statistics about the encoded data
  4. Motion compensated predictive coding, in which the differences in what has changed between an image and its preceding image are calculated and only the differences are encoded
  5. Bi-directional prediction, in which some images are predicted from the pictures immediately preceding and following the image

The first three techniques are also used in JPEG file compression.

A proposed MPEG-3 standard, intended for High Definition TV (HDTV) was merged with the MPEG-2 standard when it became apparent that the MPEG-2 standard met the HDTV requirements.

MPEG-4 is a much more ambitious standard and addresses speech and video synthesis, fractal geometry, computer visualisation and an artificial intelligence approach to reconstructing images. MPEG-4 addresses a standard way for authors to create and define the media objects in a multimedia presentation, how these can be synchronized and related to each other in transmission, and how users are to be able to interact with the media objects.

Future developments

MPEG-7 was standardised by ISO in September 2001. Formally named 'Multimedia Content Description Interface' MPEG-7 is a standard that will provide a set of tools to describe multimedia content that is passed onto, or accessed by, a device or a computer code. MPEG-7 is related to metadata, or 'content that describes the content that is shipped along with the content'. Markets for MPEG-7 will extend from the capture, production, archiving, distribution and consumption of content.

MPEG-21 will provide a larger, architectural framework for the creation and delivery of multimedia. It defines seven key elements:

Presently, the details of various parts of the MPEG-21 framework are still in draft stages.




BeoTech 2006 All rights reserved